[mythtv-users] Minimizing a Distribution

[usenet at wingert.org]

Your points are all very valid.  I don't think the point is to emulate
Redhat/Debian/Slackware.

My point would be to make a targeted distribution for a specific piece of
hardware (in my case a sufficiently powered ITX board, which we do not
have today, hopefully April).

My goal is world domination of an open sourced Linux-based PVR.  :)  I 
want to see an open sourced PVR win the PVR wars and not have to sucumb 
to mediocre hacks on some proprietary piece of hardware (TiVo/Replay/Media
Center).

I think it would be very cool for say idotpc.com to sell a mythtv based
ITX PVR COTS (ala Lindows)...  and I am sure it will be coming.  

I am sure there are many resellers waiting for this product (or FreeVo)
to mature.

IMHO, the only major component missing is a cheap hardware based MPEG2 
encoder with Linux driver support.  I think that is coming too (hopefully
the PVR-250).

Snapstream has made a pretty good business on the PVR.

Maybe I'm just dreaming.

> 
> Is trying to move MythTV from a sophisticated app to a mini-distribution 
> really a good idea? Personally, I doubt it. Consider some of the things 
> that such a distro needs to deal with:
> 
> 1. All the standard kernel-related issues regarding filesystem/partition 
> types, NIC drivers, sound-card drivers, v4l drivers, and more.
> 
> 2. All the usual X issues, including kernel framebuffers.
> 
> 3. All the TV-out issues ... and remember that many TV-out systems (Matrox, 
> ATI, nVidea) require specialized framebuffers, X drivers, &/or control apps 
> (matroxset, atitvout, nvtv). The Matrox framebuffer patches for enabling 
> TV-out are specific to kernel 2.4.19 (and 2.5.something_or_other) ... are 
> we willing to live with that kernel restriction indefinitely, or will 
> "MythOS" need to include multiple kernel choices?
> 
> 4. All the 802.11 issues, some of which require modules outside the kernel 
> (e.g., linux-wlan-ng for Prism-based cards).
> 
> 5. All the lirc issues for people using remotes.
> 
> 6. The usual distro-level issues involving updates, particularly 
> security-related patches.
> 
> 7. Incidentals like setting up networking, handling time sync'ing via ntp, 
> and providing for remote-access options (ssh? smb mounts? nfs?).
> 
> I'm probably leaving some important questions out, but this list already 
> seems big enough to give one pause.
> 
> Although I've only installed MythTV on a test basis up to now, that's 
> because my work involves addressing these other issues in the context of 
> developing a dedicated Linux-based video platform. My take on them is that 
> getting it all worked out when you can select specific target hardware to 
> develop for is bad enough. Trying to create -- AND maintain -- a distro 
> that handles all possible (or at least all plausible) hardware combinations 
> is a lot of work, and it would distract the developers here from improving 
> the core functionality of MythTV as an app.
> 
> Perhaps some of those of you who have (below) described your own 
> configurations would take a minute to consider, then discuss, how easy or 
> hard it would be to generalize your approaches to handle the variety of 
> sound cards, TV cards, video displays, NICs (including wireless), and other 
> hardware that we see discussed here regularly.
> 
> I suspect a more modest initial goal -- a FAQ-like document describing 
> approaches to setting up a host for running MythTV, as someone suggests 
> below -- is a more realistic target, at least for the short term. Included 
> here could be device-specific advice for, at a minimum, the various TV 
> cards, sound cards, and TV-out display possibilities. This alone would 
> eliminate a lot of the repeated Q&As and the "search the list archive" 
> responses I see here on this list.
> 
> 
> At 09:48 AM 2/28/2003 -0800, Larry Matter wrote:
> >At one point, there was discussion on this list to create a MythTV+distro
> >on a CD.   Not sure what happened to that.
> >
> >The idea was you could plunk a CD into a machine, reboot, and you have a
> >MythBox.  Personally, I'm not that interested in this solution.
> >
> >My personal interest is to create a disk image that a small, quiet, and
> >diskless frontend box could boot over the lan.  I'm sure everyone else has
> >slightly different needs.
> >
> >But we could at least compare notes toward these ends; like trimming down
> >the distro, best lightweight window manager, etc.  There is actually a lot
> >of this information already scattered accross the archives, but it would
> >be nice to gather it all together in a FAQ or somesuch.
> >
> >Larry
> >
> > > I am for this too.  I would love to have a small, light weight install
> > > for my  media center.  I do not have a need for a desktop attacehd to my
> > > 61 inch TV :-)
> > >
> > > My vote would be to base it on gentoo and create a "live cd" for the
> > > install.   Nothing to compile.
> > >
> > > --Ken
> > >
> > >
> > > Quoting Micah Morton <micah at oregontech.net>:
> > >
> > >> I'm voting we all get together and make a MythOS distro of linux. :)
> > >> It would be everything we want, and nothing we dont. :) Who's with
> > >> me!?
> > >>
> > >> --Micah
> > >>
> > >>
> > >> > I am in the process of booting MythTV on a Mini-ITX with Compact
> > >> Flash boot drive.
> > >> >
> > >> > I have Linux in about 20 MB (with ssh/nfs and a whole bunch of other
> > >> stuff). I have X in about another 30 MB (probably much less with a
> > >> cramfs).
> > >> >
> > >> > Boot time is really fast.
> > >> >
> > >> > The only problem I have had is minimizing QT and MySQL, they are
> > >> both nasty disk size pigs.  QT does something funky stuff loading
> > >> the MySQL libraries.
> > >> >
> > >> > The last error I am able to debug is a relocation error on a MySQL
> > >> library (I assume from QT).
> > >> >
> > >> > Has anyone seen this error?  Or better yet has anyone tried to
> > >> minimize a distribution, so I don't have to re-invent the wheel.
> > >> >
> > >> > Second question, is MySQL really necessary?  I really question the
> > >> need for a database with today's processor.  The bloat of code vs
> > >> the "time savings".  With work projects, I always wonder if a flat
> > >> text file isn't sufficent (hell we aren't cataloging Walmart's
> > >> inventory).
> > >> >
> > >> > Thanks for any pointers
> 
> 
> 
> 
> 
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at snowman.net
> http://www.snowman.net/mailman/listinfo/mythtv-users
> 
>